| DOI | Resolve DOI: https://doi.org/10.1109/ICASSP49660.2025.10888920 |
|---|
| Author | Search for: Xu, Haoning; Search for: Li, Zhaoqing; Search for: Jin, Zengrui; Search for: Wang, Huimeng; Search for: Chen, Youjun; Search for: Li, Guinan; Search for: Geng, Mengzhe1; Search for: Hu, Shujie; Search for: Deng, Jiajun; Search for: Liu, Xunying |
|---|
| Affiliation | - National Research Council of Canada. Digital Technologies
|
|---|
| Format | Text, Article |
|---|
| Conference | 2025 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP, April 6 - 11, 2025, Hyderabad, India |
|---|
| Subject | low-bit quantization; mixed-precision quantization; speech foundation model |
|---|
| Abstract | This paper presents a novel mixed-precision quantization approach for speech foundation models that tightly integrates mixed-precision learning and quantized model parameter estimation into one single model compression stage. Experiments conducted on LibriSpeech dataset with fine-tuned wav2vec2.0-base and HuBERT-large models suggest the resulting mixed-precision quantized models increased the lossless compression ratio by factors up to 1.7x and 1.9x over the respective uniform-precision and two-stage mixed-precision quantized baselines that perform precision learning and model parameters quantization in separate and disjointed stages, while incurring no statistically word error rate (WER) increase over the 32-bit full-precision models. The system compression time of wav2vec2.0-base and HuBERT-large models is reduced by up to 1.9 and 1.5 times over the two-stage mixed-precision baselines, while both produce lower WERs. The best-performing 3.5-bit mixed-precision quantized HuBERT-large model produces a lossless compression ratio of 8.6x over the 32-bit full-precision system. |
|---|
| Publication date | 2025-04-06 |
|---|
| Publisher | IEEE |
|---|
| In | |
|---|
| Language | English |
|---|
| Peer reviewed | Yes |
|---|
| Export citation | Export as RIS |
|---|
| Report a correction | Report a correction (opens in a new tab) |
|---|
| Record identifier | 56faf904-e182-46a7-b427-7135c3d72b95 |
|---|
| Record created | 2025-04-03 |
|---|
| Record modified | 2025-04-08 |
|---|
